This looks at replacing malloc/free for small temporary strings with VLAs, using a stack-first approach with heap fallback. Benchmarks were a bit surprising—allocator behavior (especially on musl) changes the trade-offs more than I expected. Interested in how others approach temporary allocation patterns.